[ Beneath the Waves ]

Thermal versus Near Infrared

article and photographs by Ben Lincoln with contributions by Dr. David Wilson

 

Perhaps the most common misunderstanding with regards to multispectral photography concerns the imaging of infrared radiation. Most of us (myself included) learn in school that "infrared" and "heat" are the same thing. Then when we learn about "infrared photography", we naturally assume that this is the same thing as thermal imaging.

Refer for a moment to the electromagnetic spectrum chart from the beginning of A Detailed Introduction. "The infrared" is commonly divided into three distinct bands, the combination of which is much larger than the entire human-visible spectrum.

When photographers refer to "infrared", they are usually describing the near infrared - the part closest to red light - generally between about 750nm and 1000nm in wavelength. This type of infrared energy behaves almost exactly like human-visible light. It can be focused using glass lenses, and the sensors used in digital cameras can easily detect it[1]. Film cameras can be used for this type of photography as well, provided that special infrared-sensitive film is used. The near infrared is also where most night vision systems and infrared remote controls operate[2].

Thermal imaging equipment generally operates in the far infrared region - waves as much as 15 times longer than near infrared light - although some types make use of mid infrared (or both far and mid infrared). Not only are specialized sensors needed, but the lenses of these cameras must be made of exotic glass (created from elements like germanium instead of the silicon which makes up everyday glass). These lenses are completely opaque to visible light, and cannot be used for standard photography.

As it is a specialized application, thermal imaging gear is very expensive compared to conventional cameras, and until about 2015 was well beyond the reach of all but the most deep-pocketed hobbyists. By 2015, it was possible to obtain an imager with a resolution of 320x240 pixels for about $4000 on the open market[3], and less than $1000 for extremely low resolution (80x60 or similar). Some of these lower-resolution models could be hacked into the equivalent of their higher-resolution counterparts if one didn't mind voiding the warranty (see Thermal Imaging Examples). Although this was about the same price a professional photographer would pay for a high-end DSLR body, these devices are much lower resolution - 320x240 pixels is less than even the cheapest bargain-basement webcam delivers in the visible spectrum, and 80x60 is not really useful for most artistic purposes. By late 2017, competition in the marketplace had increased the availability of better hardware at lower prices - for example, FLIR's VUE camera for drones supported 640x512 resolution and was available for about $3200, but this is still a far cry from the tens-of-megapixels resolution of conventional cameras.

FLIR Systems produces several relatively high-resolution models intended for advanced scientific and military use, but these are of the "call for pricing" rarefied-air budget class. They also sell thermal camera modules for smart phones. I think there is fascinating potential for that technology, but the sensor resolution in the current generation is far too low for my purposes.

The most common misconception that I've encountered over the years is the belief that a near infrared imaging system is equivalent to a thermal imager. For example, I've seen misinformed or unethical sellers on eBay offering NIR-related products with accompanying images that were clearly taken from thermal imaging gear (sometimes even with the FLIR watermark).

Dr. David Wilson came across the original version of this article and pointed out that in my attempt to clear things up, I'd actually included some technical inaccuracies from the perspective of the underlying physics. As a result, I've attempted to rewrite the relevant content in order to give three different perspectives[4] on the key question "is a near infrared imaging system a thermal imager?". Thank you, Dr. Wilson! Any mistakes which survived the rewrite are my own.

The Short Answer

No. Near infrared and what most people would think of as "thermal imaging" equipment is physically different. If one wants to see the world like the alien in Predator, one needs to use a thermal imager. There are plenty of interesting things to see in the near infrared band, but near infrared images look nothing like thermal images unless the photo is exclusively of something that is hot enough to glow incandescently.

The Medium-Length, But Still Simplified Answer

Near infrared equipment and what most people would think of as a "thermal imager" are physically different, and each captures a distinct, non-overlapping part of the electromagnetic spectrum. If one wants to see the world like the alien in Predator, one needs to use a thermal imager. There are plenty of interesting things to see in the near infrared band, but near infrared images generally look nothing like thermal images.

The exception is images of objects which are hot enough to glow incandescently in the near infrared.

All matter with a temperature above absolute zero produces electromagnetic waves - "thermal radiation" - as a result of atomic-level processes. The wavelength of this radiation is extremely long (radio/microwave) for very cold matter, but the minimum wavelength emitted decreases as the temperature rises. For most objects and environments a typical human encounters, thermal radiation includes a significant amount of far- and mid-infrared. It is that far- and/or mid-infrared which a thermal imager captures.

As the temperature of matter rises, the thermal radiation will extend to shorter wavelengths, eventually reaching the near infrared, then the human-visible spectrum, and then to shorter wavelengths. We perceive this as incandescent light. The glow of a hot stove element is caused by the same physics as the surreal glow of a human as captured by a thermal imager.

As a result, if an object is hot enough, it can be argued that a near infrared (or even conventional) camera can function as a "thermal imager", because it is capturing the emissions that result from the heat of the object. However, this is only true for objects that are far too hot to touch, and differentiating between "incandescent light" and "some other source of the same spectral emissions" is much harder than for mid- and far-infrared. Numerous objects produce near infrared, red, orange, yellow, and white light without being as hot as an incandescent object of the same colour. Far fewer produce far- and mid-infrared radiation that would confuse a thermal imager.

The (Hopefully) Technically Correct and Mostly-Complete Answer

Just about everything you probably think you know about heat is a misleading oversimplification based on the assumption that you will spend your life in "normal" environments and not study physics.

Many people (myself included, before I researched this article) believe that "heat" is equivalent to some form of infrared radiation. This is not true, although for most situations humans encounter, treating it as true is a more-or-less accurate simplification.

In reality, heat (in the sense of electromagnetic waves emitted by an object based on its temperature - "thermal radiation") can take the form of many different wavelengths. It is produced by oscillations at the atomic and molecular level. This occurs for any matter above absolute zero in temperature (because at absolute zero, all motion would cease, preventing any oscillations).

The wavelengths produced at a given temperature can be approximated using the black-body radiation model (see previous link for more details). A given temperature of matter will produce EM radiation with wavelengths occupying a curve. For very cold objects, this curve begins in the microwave or radio band. As the temperature rises closer to values that might be encountered outside of a laboratory, shorter wavelengths are also emitted: far infrared, mid infrared, near infrared, light that is visible to humans, and so on.

I used a handy Excel workbook created by Scott Sinex and EasyCalculation.com's Wien's Displacement Law Calculator to derive the data on this page.

Modern thermal imagers typically have a minimum detection temperature of -20 degrees Celsius. This corresponds with the point at which the blackbody radiation curve's emissive power exceeds ((10 watts/meter^2) * (wavelength in micrometers)), as well as the point at which the peak wavelength is about 11,500nm, but either or both of these may be coincidences. Temperatures far lower than this will still produce decent amounts of far infrared at longer wavelengths, so even if one had not already learned that "heat" and "infrared" are not the same thing, it would still be obvious that a "thermal imager" is actually more accurately termed "an imager of thermal radiation produced in a fairly specific range of temperatures".

When the temperature reaches high enough levels, the thermal radiation will be of sufficiently short wavelengths to be imaged using near infrared, or even conventional cameras. This is the incandescence that is familar to anyone who has looked at a hot stove element. It is caused by the same physics as longer-wavelength thermal emissions. I would argue that treating such devices as "thermal imagers" is misleading for most environments that most people will encounter, because essentially none of the unusual/surprising aspects of thermal imagery apply, but I can see arguments on both sides.

In any case, If one wants to see the world like the alien in Predator, one needs to use a thermal imager. There are plenty of interesting things to see in the near infrared band, but near infrared images generally look nothing like thermal images.

Some statistics for various temperatures of interest:

To illustrate this discussion, here's a set of photos I took of my soldering iron while it was cold:

Soldering Iron at Room Temperature
[ R-G-B ]
R-G-B
[ NIR-Grey ]
NIR-Grey
[ UVA-Grey ]
UVA-Grey
   

This shot doesn't show any interesting variation in false colour, so I've only included the visible light and greyscale NIR/UVA versions.

Date Shot: 2009-09-09
Camera Body: Nikon D70 (Modified)
Lens: Nikon Series E 100mm
Filters: Standard Set
Date Processed: 2009-09-09
Version: 1.0

 

Next, the same soldering iron after it has heated to operating temperature, as it appears in the near infrared:

Soldering Iron at Operating Temperature
[ NIR-Grey ]
NIR-Grey
       

Look, it's hot, there's a wisp of smoke coming off the solder I put on the tip.

Date Shot: 2009-09-09
Camera Body: Nikon D70 (Modified)
Lens: Nikon Series E 100mm
Filters: Standard Set
Date Processed: 2009-09-09
Version: 1.0

 

It's hot enough to melt solder, but there is no trace of a glow visible in the near infrared, at least in this case. Apparently, it's possible to capture an NIR glow from a soldering iron in a dark room, but I have not succeeded at this to date.

Here is a similar soldering iron warming up as captured using a thermal imager:

Soldering Iron at Operating Temperature - Thermal IR
[   ]
 
       

The appearance of a hot soldering iron using a thermal imager.

Date Shot: 2014-08-14
Camera Body: FLIR E4 (Modified)
Lens: FLIR E4
Filters: None
Date Processed: 2014-08-14
Version: 1.0

 

As discussed above, the near infrared can be used to detect one particular type of hot object: those that are almost hot enough to begin glowing incandescencently in the NIR band, which occurs at slightly lower temperatures than those at which the same object would glow red.

Here is a closeup of one of the heating elements on my electric stove, with the power turned off:

Stovetop Bokeh (Stove Off)
[ R-G-B ]
R-G-B
[ NIR-Grey ]
NIR-Grey
[ UVA-Grey ]
UVA-Grey
[ NIR-R-G ]
NIR-R-G
[ G-B-UVA ]
G-B-UVA

 

Date Shot: 2009-09-09
Camera Body: Nikon D70 (Modified)
Lens: Nikon Series E 100mm on an extension ring
Filters: Standard Set
Date Processed: 2009-09-09
Version: 1.0

 

Here is the same element at low heat in the near infrared. You can see that it is still just as dark as with no power applied.

Stovetop Bokeh (Stove on Low)
[ NIR-Grey ]
NIR-Grey
       

 

Date Shot: 2009-09-09
Camera Body: Nikon D70 (Modified)
Lens: Nikon Series E 100mm on an extension ring
Filters: Standard Set
Date Processed: 2009-09-09
Version: 1.0

 

The same element at medium heat still is not emitting any near infrared light, even though it has visibly expanded from the higher temperature, and the heat being emitting can easily be felt from 20-30cm away.

Stovetop Bokeh (Stove on Medium)
[ NIR-Grey ]
NIR-Grey
       

 

Date Shot: 2009-09-09
Camera Body: Nikon D70 (Modified)
Lens: Nikon Series E 100mm on an extension ring
Filters: Standard Set
Date Processed: 2009-09-09
Version: 1.0

 

At the "medium high" setting, finally a change emerges in the near infrared, even though the element is still dark to human eyes. The change is easiest to see in the NIR-R-G false colour rendition.

Stovetop Bokeh (Stove on Medium-High)
[ R-G-B ]
R-G-B
[ NIR-Grey ]
NIR-Grey
[ UVA-Grey ]
UVA-Grey
[ NIR-R-G ]
NIR-R-G
[ G-B-UVA ]
G-B-UVA

 

Date Shot: 2009-09-09
Camera Body: Nikon D70 (Modified)
Lens: Nikon Series E 100mm on an extension ring
Filters: Standard Set
Date Processed: 2009-09-09
Version: 1.0

 

Halfway in-between "medium high" and high, the element still is not emitting any human-visible light, but there is no mistaking that a significant near infrared glow is present.

Stovetop Bokeh (Medium-High Plus)
[ R-G-B ]
R-G-B
[ NIR-Grey ]
NIR-Grey
[ UVA-Grey ]
UVA-Grey
[ NIR-R-G ]
NIR-R-G
[ G-B-UVA ]
G-B-UVA

This is with the stove halfway in-between the "Medium-High" and "High" marks.

Date Shot: 2009-09-09
Camera Body: Nikon D70 (Modified)
Lens: Nikon Series E 100mm on an extension ring
Filters: Standard Set
Date Processed: 2009-09-09
Version: 1.0

 

At its highest setting, the element is begining to glow a dull red colour, but is putting out a stunning amount of near infrared light.

Stovetop Bokeh (Stove on High)
[ R-G-B ]
R-G-B
[ NIR-Grey ]
NIR-Grey
[ UVA-Grey ]
UVA-Grey
[ NIR-R-G ]
NIR-R-G
[ G-B-UVA ]
G-B-UVA
[ B&W 403 Only ]
B&W 403 Only
[ B&W 403 and LDP CC1 Stack ]
B&W 403 and LDP CC1 Stack
     

I've included two alternate versions of this shot to make a point about ultraviolet bandpass filters. The first is the result when just a B&W 403 filter is used, and the second is with an LDP CC1 IR-blocking filter stacked on top of it. You can see that without the IR filter, the result is not an ultraviolet image at all, but near infrared. Even with the IR filter, there is still some NIR leakage as evidenced by the glow of the stove element in that version. Compare this to the "regular" UVA version (shot using a Baader U-Filter), where the element is completely dark even though it is blazing away in the NIR.

Date Shot: 2009-09-09
Camera Body: Nikon D70 (Modified)
Lens: Nikon Series E 100mm on an extension ring
Filters: Standard Set plus B&W 403
Date Processed: 2009-09-09
Version: 1.0

 

This last set reinforces my belief that false colour is at least generally preferable to discreet greyscale images. The greyscale near infrared version indicates to the viewer that something is very bright. The false colour NIR-R-G variation intuitively depicts that the stove element is very hot. If you were wearing a set of multispectral goggles, which would you rather see before deciding whether to touch the element or not?

 
Footnotes
1. See Cameras.
2. See Uses of Multispectral Photography
3. Older thermal imaging equipment can sometimes be purchased secondhand at lower prices. If you want to go this route, keep in mind that some models require active cooling via an attached container of gas, and may not offer anything other than analogue video output.
4. As the Vorlons say, "understanding is a three-edged sword".
5. The curve for temperatures this low is smooth enough that I'm a little loathe to refer to a "peak", but have included it for consistency.
 
[ Page Icon ]